💸 Migrant remittances

How much the flows weigh: who sends, who receives

Migrant remittances are the money that people living abroad send to their home country, usually to family. They are the largest private financial flow to developing countries. Low- and middle-income countries received $685 billion in 2024, up 5.8% from a year earlier. That is more than foreign direct investment and official development assistance combined (World Bank, Migration and Development Brief, December 2024). The World Bank projected 2.8% growth in 2025, bringing the total to about $690 billion. For the payments industry, this is a retail market with small average tickets, monthly frequency, and extreme price sensitivity. Half of the volume lands in instruments that are not bank accounts.

The relative weight of remittances compares what a country receives with its gross domestic product. The absolute amount measures the size of the market. This ratio measures how politically sensitive the flow is, and therefore the regulatory risk an operator takes on. Remittances equal about 45% of GDP in Tajikistan, 38% in Tonga, and 27% in Nicaragua and Lebanon alike (same brief, 2024 data). In these economies, a money transfer operator is part of the macroeconomic infrastructure, and the central bank supervises it as such. It is exposed to decisions on exchange rates, price caps, or licensing that have no equivalent in the card market.

Two other markets show how far regions diverged. The Philippines posted a record $35.634 billion in cash remittances in 2025, up 3.3% from $34.493 billion in 2024 (Bangko Sentral ng Pilipinas, February 2026). The US was the largest declared source at 39.7%, ahead of Singapore (7.3%) and Saudi Arabia (6.6%). Nigeria held flat at $21.806 billion in 2025, against $21.811 billion in 2024 (Central Bank of Nigeria), after an overhaul of the International Money Transfer Operators framework designed to pull flows back into formal channels. Mexico’s decline, the Philippine record, and Nigeria’s plateau all happened in the same year. Each corridor follows its own growth path, and the global average blends them without describing any of them.

Anatomy of a transfer: what really happens between sender and recipient

A remittance usually runs as two separate domestic transactions, linked by deferred settlement between the operator’s entities. In the vast majority of cases, nothing crosses the border when the transaction takes place. The operator collects funds in the sending country and pays out in the receiving country from funds it placed there in advance. It offsets the two legs later, in bulk. Wise works this way with its local accounts in each country, Western Union with its agents, and hawala in its own way. This setup determines where cost, risk, and capital requirements sit, as the steps below show one by one.

Offer A - "4.99 USD fee"             Offer B - "no fees"
------------------------------       ------------------------------
Amount sent         200.00 USD       Amount sent         200.00 USD
Displayed fee         4.99 USD       Displayed fee         0.00 USD
Interbank rate      1 USD = 17.00    Interbank rate      1 USD = 17.00
Applied rate        1 USD = 16.83    Applied rate        1 USD = 16.32
FX margin               1.00%        FX margin               4.00%
------------------------------       ------------------------------
Total cost            6.99 USD       Total cost            8.00 USD
i.e.                    3.50%        i.e.                    4.00%

Same amount received? NO.
Offer A: (200 - 4.99) x 16.83 = 3,282 local units
Offer B:  200         x 16.32 = 3,264 local units

The only honest metric is the AMOUNT RECEIVED by the recipient,
for a given amount sent, on a given date.

The operators: agent networks, digital natives, aggregators

The remittance market has three layers, stacked from the commercial front end down to the physical roots of the service. The first is the consumer brands the sender sees. The second is the payout aggregators, which sell white-label access to local rails and which the end customer never sees. The third is the physical distribution networks: agents, shops, grocery stores, and post offices. The value of that last layer cannot be replicated in software. A fintech can copy an app in a year, but a network of 500,000 payout locations is built one merchant at a time.

The aggregator layer determines how far an operator’s geographic reach can extend. Onafriq (formerly MFS Africa) claims 43 African countries connected, 1 billion mobile wallets reachable, and 2,000 cross-border corridors. These figures are self-reported and unaudited, and should be read that way. Its structural value is that a single integration replaces dozens of bilateral agreements, corridor by corridor. Nium, which grew out of InstaReM, claims more than $60 billion in annual volume and its own licenses in more than 40 markets, which sets it apart from platforms that rent a partner’s license. Thunes and Airwallex play similar roles in other segments. The sending operator then faces a choice. It can carry the regulatory risk itself in 40 countries, or buy that coverage from an aggregator that already carries it, in exchange for a fee on every payout.

Measured cost: what SDG 10.c really requires

Sustainable Development Goal target 10.c commits countries to reducing the transaction cost of migrant remittances to below 3% by 2030. By the same date, it also calls for eliminating corridors that cost more than 5%. The tracking indicator, 10.c.1, draws on the World Bank’s Remittance Prices Worldwide database, published quarterly and covering 365 corridors, from 48 sending countries to 105 receiving countries. It is the only price benchmark the industry can be held to. Any offer billed as “cheaper than the market” is implicitly measured against the averages this database publishes for its corridor.

The most revealing cost gap lies in the instruments used, more than in the receiving regions. The RPW database distinguishes four types of provider: banks, money transfer operators (MTOs), mobile operators, and post offices. Banks came in at about 12% in the fourth quarter of 2023, the most expensive channel in the world, while mobile operators came in at 4.4%, the cheapest. Yet mobile operators then handled less than 1% of the volume transferred. The cheapest channel was already available and working, but it captured no meaningful share. That gap between the cheapest technology and actual distribution is the industry’s central problem.

These figures call for a methodological caveat. The RPW global average goes up as often as it goes down. It rose from 6.2% to 6.7% between the second quarter of 2023 and the second quarter of 2024 (UN Sustainable Development Goals Report 2025, citing RPW data). It then eased to 6.49% and then 6.36% in 2025. These moves reflect changes in the sample (corridors dropped, corridors added) as much as real price cuts. A shift of one or two tenths of a point in a quarter therefore cannot distinguish a price cut from a sample change. The trends worth tracking in this database are the SmaRT average and the share of corridors above 5%.

The last mile: mobile money, wallets, and agent cash-out

The last mile is the step in which the recipient actually takes possession of the funds at home. On the main corridors, it runs through an e-money wallet or cash pickup, and rarely through a bank account. Mobile money passed $2 trillion in transactions in 2025, up 23% from a year earlier (GSMA, State of the Industry Report on Mobile Money, 2026). That total has doubled in four years. International remittances are only a small share of it, which the GSMA estimated at about 4% of global remittances in 2023. That share is growing, and it is structurally cheaper. Sending via mobile money costs about 44% less than the global all-channel average. More than 55% of services offering cross-border mobile money transfers are below the 3% SDG target, up from 43.5% in 2022.

Cash withdrawal remains the core step on many corridors, and the least understood link. A cash-out agent does not run a counter stocked by a bank: it is a merchant fronting its own cash. The agent must hold both an electronic balance (to accept deposits) and a stock of banknotes (to serve withdrawals). The balance between the two erodes as soon as a neighborhood becomes a net receiver, which is exactly what happens in areas that receive remittances. The operator must then arrange physical or electronic rebalancing, which has a real cost that feeds into the commission. The GSMA tracks how much agents convert into e-money. Agents digitized $430 billion in 2025, about 20% more than a year earlier, and more than all other inflows (bank transfers, bulk payments, and international remittances) combined.

• Agent liquidity. A corridor can be technically flawless and still fail because the village agent has no banknotes on the 3rd of the month. The service rate at peak hours is an operating metric, not a detail.
• Agent commission. It is charged on the withdrawal, it is often invisible in the price the sender compares, and it can add several percentage points on small amounts.
• Recipient identification. A phone number is not an identity. Keying errors credit the wrong person, and the funds are all but unrecoverable on rails with irrevocable push credits.
• Interoperability. Sending to wallet A when the recipient uses wallet B requires an interconnection agreement. Without one, the recipient withdraws cash and redeposits it elsewhere, paying twice.
• Regulatory limits. Most e-money regimes impose balance and monthly transaction limits tied to the level of identity verification. A lightly verified recipient can have a payment rejected without any system being at fault.

When governments build the rails: lessons from Directo a México to UPI–PayNow

A public remittance rail is payment infrastructure run by central banks or public operators and opened to transfers between two countries. The rationale is old. If remittances are expensive because private intermediaries take a margin, connecting the two countries’ public infrastructures directly should fix the problem. Twenty years of experience show that the rail is never enough. The plumbing works, but commercial distribution, awareness among senders, and the last mile still have to be built.

Directo a México is the textbook case of a public remittance rail. It has run since 2003, uses the central bank exchange rate, and costs a fraction of what MTOs charge. Yet it has never captured meaningful volume. The reason is the sender’s profile, not a technical limit of the rail. A migrant worker pays cash, often without a US bank account, and goes to the neighborhood store where the staff speak the sender’s language. An interbank rail remains out of reach for that sender as long as no physical entry point leads to it. The same mechanism works in reverse for links between instant payment systems. UPI–PayNow works because both countries already have a user base on their domestic systems and a fully digital last mile on both sides.

In Southern Africa, TCIB (Transactions Cleared on an Immediate Basis), run by PayInc since 2021, has a different aim. It formalizes corridors that are still largely informal today, from South Africa to Zimbabwe, Malawi, and Mozambique. Its goal is to bring into the measured system flows that now travel by bus, by suitcase, and through networks of trust, rather than to lower the advertised price. The Gulf follows the same logic with AFAQ, which settles same-day in six local currencies between the RTGS systems of the six GCC countries. Buna is the only regional rail that reaches beyond the Gulf to the Levant and North Africa.

Compliance, licensing, and de-risking: what breaks a corridor

In this market, compliance is not a back-office function. It is the largest variable cost after FX, and the leading reason corridors close. Two forces pull in opposite directions. On one side, traceability requirements are rising. FATF Recommendation 16, revised in June 2025, requires originator and beneficiary data to travel with the payment end to end, in a form that is as structured as possible, with full implementation expected by end-2030. On the other, correspondent banks are pulling back. The number of active correspondent relationships fell by about 30% between 2011 and 2022. The number of active corridors dropped from nearly 10,800 to 9,800 between 2011 and 2018, even as the value of cross-border payments grew (BIS, Bulletin no. 87, May 30, 2024).

• Sanctions screening and name matches. A name transliterated from a non-Latin script triggers an alert and a manual review. On a $200 ticket, two minutes of analyst time wipe out the transaction’s margin.
• Thin recipient data. A PO box, a “c/o,” a bare phone number: the revised Recommendation 16 pushes toward structured fields (address, date of birth, legal entity identifier) above a threshold of USD/EUR 1,000.
• Bank de-risking. A nonbank PSP can have its account closed purely because of its bank’s risk policy, with no incident or sanction. This is the leading cause of sudden corridor shutdowns, and it cannot be challenged in court.
• Intermediary concentration. On a fragile corridor, only one or two correspondents are willing to carry the local leg. They set the price and the timing, and decide on their own when to stop.
• Exchange controls. In several receiving markets, conversion into local currency is administered. The rate applied to the recipient is not the market rate, and the gap shows up in no price comparison tool.

Stablecoins: a settlement rail first, a payment method second

A stablecoin is a digital token whose value is pegged to a reference currency, usually the US dollar. In remittances, it is currently used on only one leg of the transaction: settlement between the operator’s entities, step seven in the flow described above. The recipient still receives local currency, from an agent or in a wallet. The shift is real, measurable, and already live at leading players. It affects prefunding and the correspondent bank, without changing the last mile.

The legal framework for stablecoins came together in 18 months, and it now shapes the choice of a settlement token. In the US, the GENIUS Act (2025) sets up a federal framework for payment stablecoins, with supervision by the federal banking regulators and by qualifying state regimes. In the EU, MiCA (Titles III and IV) governs e-money tokens and asset-referenced tokens, with national authorization and enhanced EBA supervision of significant issuers. In Hong Kong, the Stablecoins Ordinance (Cap. 656), passed on May 21, 2025, requires an HKMA license to issue. The three regimes converge on one practical requirement. A remittance operator that settles in stablecoins must now choose an issuer licensed in the jurisdiction where it operates, because a token’s liquidity alone no longer justifies using it.

Opening a corridor: a practitioner’s checklist

Opening a remittance corridor requires four conditions, and all of them must hold at the same time: the right to operate at both ends, access to local liquidity, a last mile the actual recipient can reach, and a competitive price on the only metric that counts, the amount received. None of them is solved just by switching on a payment method with a PSP.

• The right to operate, on both sides. A money transmission or e-money license in the sending country, and licensed status or a licensed partner in the receiving country. In the US, the process runs state by state; in West Africa, e-money issuer licenses are granted country by country despite the monetary union.
• The local leg and how it is funded. Three parameters are set together: who holds the payout account, how many days of flows to keep there, and the intraday FX risk on that balance. On a corridor with a nonconvertible currency, this funding alone determines whether the economics work.
• Where the recipient actually receives. Not the default bank account, and not the best-known wallet, but the one this diaspora segment uses. A Gulf→Bangladesh corridor ends at bKash, Nagad, or Rocket, not at a commercial bank.
• Interoperability and cash-out. The key question is whether the recipient can use the funds without withdrawing them. If not, the corridor depends entirely on the agent network, on agent commissions, and on their service rate at month-end. That is where corridors that worked in testing fail.
• Price as the amount received. The metric to publish and track is the amount received for $200 sent, rate and fees included, compared with the corridor’s SmaRT average in the Remittance Prices Worldwide database. Any other framing lets the operator choose which part of the price to show: the displayed fee on one side, the FX margin on the other.
• Compliance sized for the average ticket. At $200 per transaction, manual review is not sustainable. Screening, false-positive handling, and structured data collection must be designed for a unit cost of a few cents, not a few euros.
• A banking fallback plan. Line up a second settlement partner in the receiving country as soon as the corridor opens. De-risking is not announced in advance; it is notified, with contractual notice of 30 to 90 days.